Testing umbrella species and food-web properties of large carnivores in the Rocky Mountains

Despite criticisms, the umbrella species concept remains a fundamental conservation tool for protecting biodiversity in the face of global change, yet it is rarely tested. Food web theory provides a tool to test both umbrella-species' suitability and their ecological function, which we investigate in a large-mammal food web. Using data from 698 camera trap locations in the Canadian Rockies, we develop hierarchical occupancy models to predict the co-occurrence of 16 large mammal species. We draw upon previous diet studies in the Canadian Rockies to describe the meta food-web (meta-web) for these species. Next, we filtered the meta-web using predicted occupancy to estimate realized food webs at each camera location. We tested the umbrella species concept using predicted occupancy across all 698 camera sites. We then tested for carnivore effects using realized food webs on 5 food-web properties: species richness, links, connectance, nestedness and modularity using generalized linear models while accounting for landscape covariates known to affect food web dynamics. Our multispecies occupancy models reflected factors previously demonstrated to affect large mammal occurrence. Our results also demonstrated that grizzly bear (Ursus horribilis), a generalist carnivore, was the best umbrella carivore species, and explained species richness the best. When considering food web properties, however, wolves (Canis lupus) and cougars (Felis concolor) served as better umbrellas that also captured food web properties such as connectance, links and nestedness that better reflect ecological interactions. Our results support the role of large carnivores as umbrella and ecologically interactive species in conservation planning.